3D dense reconstruction based on omnidirectional camera and laser range finder

To produce the 360-degree three dimensional reconstruction, a calibration and data fusion methods were proposed based on omnidirectional camera and laser range finder. Using the 2D checkerboard for joint calibration, our method doesn't rely on the edge information of the lidar data, thus avoids the error of feature extraction. We also proposed a dense 3D reconstruction method based on superpixel. First, an image is segmented into superpixel blocks, then the plane of the block is estimated based on the lidar data, and last the depth information of each pixel in the superpixel block is calculated. As the experimental results show, the calibration is accurate, the contour of the scene is clear and the mapping of the color is correct.